From the numerical point of view the Large Eddy Simulation can be interpreted as a procedure to solve under resolved Navier-Stokes equations. As such the methods applied to model the scales smaller than the scale resolved on the computational mesh can be read as one of the many ways to increase the numerical resolution.From the theoretical point of view a Large Eddy Simulation can be formalized as a filtered solution of the Navier-Stokes equations. Filtering can be done in different ways : truncations in the spectral space, convolutional integrations in the physical space and in general we can formalize all that operationally as due to some averaging operator provided with some particular properties. Obviously the theoretical difficulties are due to the fact that it is relatively easy to filter data but usually very hard to filter equations.In these lectures the fundamentals of the Large Eddy Simulation will be examined and discussed in order to provide and as possible to justify the main ingredients of such approach, the subgrid scale models. The practical objectives of the Large Eddy Simulations and how they are located between the Direct Numerical Simulations and the Reynolds Averaged Navier Stokes solutions will be exposed. Particular emphasis will be given to the operational approach and to its principal product, the dynamic modelling procedure.
et al. (eds.), Advanced Turbulent Flow Computations